Spectral Domain Decomposition Using Local Fourier Basis: Application to Ultrasound Simulation on a Cluster of GPUs

Abstract

The simulation of ultrasound wave propagation through biological tissue has a wide range of practical applications. However, large grid sizes are generally needed to capture the phenomena of interest. Here, a novel approach to reduce the computational complexity is presented. The model uses an accelerated k-space pseudospectral method which enables more than one hundred GPUs to be exploited to solve problems with more than 3 ×10 9 grid points. The classic communication bottleneck of Fourier spectral methods, all-to-all global data exchange, is overcome by the application of domain decomposition using local Fourier basis. Compared to global domain decomposition, for a grid size of 1536 × 1024 × 2048, this reduces the simulation time by a factor of 7.5 and the simulation cost by a factor of 3.8.

High resolution 3D ultrasonic breast imaging by time-domain full waveform inversion

“…• The methods presented here were exclusively implemented in Matlab, using the GPU support offered by the parallel computing toolbox. Implementing it using optimized CUDA code will lead to further performance gains [69,83,84]. • While we presented the material in terms of a general acoustic parameter u, we only showcased the methods for u = c 0 .…”

“…In the TR based gradient computation described in section 3.1, the adjoint and TR wave simulation can be run on two separate GPUs in parallel, which would lower the computational cost to that of the standard gradient computation again. To reconstruct 3D images on higher resolutions, one can implement 3D domain decomposition methods to distribute the computations over several GPUs [69,83,84]. A straightforward way which does not need sophisticated implementation is to average statistically independent gradient estimates each computed on a different GPU.…”